16 July 2009

Earthquake induced landslides

Posted by Dave Petley

Every so often an event occurs to shake up out of our complacency. So often this is a disaster that is mind-bogglingly destructive, such as the Wenchuan Earthquake landslides last year. However, just occasionally something that happens that is far less damaging than would be expected. This of course is easier to ignore, but in fact can be just as informative as the big events.

Yesterday is just such a case. Sitting in a meeting I received a GDACS alert to say that there had been a Magnitude 8.2 earthquake in the far south of New Zealand. The earthquake was shallow – less than 20 km – and the south of New Zealand is highly mountainous, suggesting that it was likely to have induced a large number of landslides. The last big event in this area, the Fiordland earthquake of 21st August 2003, which was”only” Mw=7.2, triggered lots of slides (see the excellent Geonet report on this event), so the assumption that this large event would do the same was quite reasonable. Since the event the earthquake has been downgraded by the USGS to Mw=7.6 at a depth of 12 km – although smaller I would expect that this would still be a massive landslide-inducing event.

I’m wrong. In fact report coming from fly-over surveys of the epicentral area suggest that there were very few landslides triggered by the event. This is really surprising. Over the last 25 years, since the pioneering work by David Keefer at the USGS, a lot of work has been done examining the relationship between landslides and earthquakes. The map below shows the distribution of well-documented studies of earthquake-induced landsliding apologies if I missed out your particular study – please let me know!) – each yellow dot is an earthquake event for which extensive landslides have been documented. The background image is the GSHAP earthquake hazard map – dark areas have a high level of hazard:


The colours indicate the size of the earthquake and the dots are located at the epicentre of the earthquake (which is why some are obviously offshore). You will see that there is a pretty good coverage of areas that are obviously both seismically hazardous and have high relief, with some areas of high concentration because of proximity to research teams (Italy and California for example). You will also see that New Zealand, thanks to the efforts of GNS and Mike Crozier at Victoria University in Wellington, is pretty well covered.

For a substantial proportion of these earthquakes the area affected by landslides has been measured. Unsurprisingly, there is a pretty strong relationship between the area affected by landslides and the earthquake magnitude:

So when an earthquake occurs in a mountainous area we have a pretty good idea of the area that we would expect to be affected by landslides. Note that there is quite a large range for any given earthquake magnitude – this is the influence of earthquake depth, topography, prior weather conditions (i.e. has it been wet, in which case the ground is likely to be less stable), vegetation, humans, etc.

This map suggests that we should expect to see many landslides for the event yesterday. This is clearly not the case – which is something of a surprise. It will be interesting to see what happened here – for that we will have to wait for more detailed studies over the next few months. I wonder if the dynamics of this event might be slightly unusual, accounting for the vast difference in initial estimates of magnitude between the USGS and Geonet? Perhaps this translated into much lower ground accelerations than might be expected – and hence the low number of landslides.